Post your SiSoft Sandra Storage Results

[playafly187]

Post your SiSoft Sandra HDD scores here. To make this sticky as useful as possible, please keep questions and comments to PM's or other threads only. If can download the latest version of Sandra here:

http://www.ocia.net/sandra/sandra2005.exe

To verify your scores, please post a screenshot of your results if possible. Also, we ask that you post your complete system specs in the following format:

CPU type and clock speed:
Cooling:
Memory (brand, capacity, timings, voltage, etc):
Motherboard:
Hard Drives:
Sandra Version Number:

[playafly187]

A64 3000+ Venice @ 300 x 9 1.65v
XP90 HSF
DFI NF4 Ultra-D
2x 512mb ocz pc4800 platinum elite @ 2.5-4-4-10 2T 2.8v
74GB WD Raptor 10K RPM HDD
Sandra 2005.1.10.37

http://www.ocia.net/playafly187/hddforum.jpg

[FunkZ]

A64 3000+ Venice @ 2.9
XP-90
Chaintech VNF4
2 x 256 Geil Ultra PC4400 3-4-4-8-2T 2.8v
2 x 74GB Raptor RAID0
Sandra 2005.1.10.37
HD Tune 2.10

http://mywebpages.comcast.net/funkz/raptor74raid0ss.jpg
http://mywebpages.comcast.net/funkz/raptor74raid0hdt.jpg

[jester]

Version: 2005.7.10.60

2 Hitachi Deskstar 80g sata 2 (running in sata1) 8m buffer in raid 0

http://gixer.org/image/ocia/hd/2deskstar-raid0.gif

[jester]

Same as above but with 3 in raid 0

Version: 2007.5.10.98

http://gixer.org/image/ocia/hd/3raid0.gif

The sandra version I used on the last bench gave me the same results.

[playafly187]

thought i had posted results from my 800gb raid array.. guess not:

http://www.ocia.net/playafly187/raidhddbench1.jpg

http://www.ocia.net/playafly187/raidhddbench2.jpg

[FunkZ]

Server 3800+ X2 @ 2.7
Highpoint RocketRAID 1640
3 x 160Gb Seagate 7200.7 (SATA150) RAID5

Not quite the performance of a stripe set, but not bad for a fault-tolerant array either. I think the PCI bus may be bottlenecking the throughput but unfortunately this is on an nForce4 board so only 2 of the onboard SATA ports are really available for use.

http://mywebpages.comcast.net/funkz/hpt1640raid5a.jpg
http://mywebpages.comcast.net/funkz/hpt1640raid5.jpg

Here is the result from just one of the 160Gb Seagate 7200.7 as a standalone drive for comparison.

http://mywebpages.comcast.net/funkz/seagate7200.7ss.jpg
http://mywebpages.comcast.net/funkz/seagate7200.7hdt.jpg

[FunkZ]

HD Tune from my new 3 Raptor stripe set. I spent some time testing 16k, 32k and 64k stripe sizes and 32k ended up being the best performing.

http://mywebpages.comcast.net/funkz/3raptors32kstripe.jpg

[Avalanche]

Nice. Now I wish I can change mine to 32K without redoing my raid. Anyways here is mine @ 64K (Raid 0 15.08 drivers).

http://i54.photobucket.com/albums/g81/KYA-Avalanche/Harddrivetest.jpg

[FunkZ]

From my observations, for a 2 drive stripe set 64k is the best to use and 32k for 3. The consensus seems to be 16k for 4 although I haven't tested that personally.

You want to go small enough so that when you write a file it gets broken up and distributed across (ideally) all of your drives in the array, without being too small that it causes more work for the controller to break it up and then reassemble on reads.

If you use a 128k block size for example and then save a 128k file only 1 drive gets the data and you get no speed benefit. On the flip side if you use a 16k block size the controller has to break up that file into 8 pieces which is overkill for a 2-drive array.

So the above numbers seem to work well for general use. If you're constantly dealing with much larger files like pictures, movies, etc. then larger block sizes may be more beneficial.

[Avalanche]

Thanks for the info. It explains alot.

[FunkZ]

Latest HD Tune benchmark

3 x 74Gb Raptors RAID0 32k stripe
http://home.comcast.net/~funkz/3raptors32k.jpg